International Journal of Pharmaceutics 468 (2014) 55–63

Preliminary pharmaceutical development of antimalarial–antibioticcotherapy as a pre-referral paediatric treatment of fever in malariaendemic areas

Alexandra Gaubert a, Tina Kauss a,*, Mathieu Marchivie b, Boubakar B. Ba a,Martine Lembege c, Fawaz Fawaz a, Jean-Michel Boiron d, Xavier Lafarge d,Niklas Lindegardh e, Jean-Louis Fabre f, Nicholas J. White e,g, Piero L. Olliaro g,h,Pascal Millet a, Luc Grislain i, Karen Gaudin a

aUniversité de Bordeaux, EA 4575 Analytical and Pharmaceutical Developments Applied to Neglected Diseases and Counterfeits, Bordeaux, FrancebUniversité de Bordeaux, FRE 3396 CNRS Pharmacochimie, Bordeaux, FrancecUniversité de Bordeaux, Laboratory of Organic and Therapeutic Chemistry, Pharmacochimie, Bordeaux, Franced EFS (Etablissement Français du Sang) Aquitaine Limousin, Bordeaux, Francee Faculty of Tropical Medicine, Mahidol University, Bangkok, ThailandfOTECI (Office Technique d’Etude et de Coopération Internationale), Paris, FrancegCentre for Tropical Medicine, Nuffield Department of Medicine, University of Oxford, UKhUNICEF/UNDP/WB/WHO Special Program for Research & Training in Tropical Diseases (TDR), Geneva, SwitzerlandiUniversité de Bordeaux, Laboratory of Industrial Pharmaceutical Technology, Bordeaux, France

A R T I C L E I N F O

Article history:Received 24 February 2014Accepted 8 April 2014Available online 13 April 2014

Keywords:ArtemetherAzithromycinMalariaAcute respiratory infectionsPediatricRectal route

A B S T R A C T

Artemether (AM) plus azithromycin (AZ) rectal co-formulations were studied to provide pre-referraltreatment for children with severe febrile illnesses in malaria-endemic areas. The target profile requiredthat such product should be cheap, easy to administer by non-medically qualified persons, rapidlyeffective against both malaria and bacterial infections. Analytical and pharmacotechnical development,followed by in vitro and in vivo evaluation, were conducted for various AMAZ coformulations. Of theformulations tested, stability was highest for dry solid forms and bioavailability for hard gelatin capsules;AM release from AMAZ rectodispersible tablet was suboptimal due to a modification of its micro-crystalline structure.ã 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license

(http://creativecommons.org/licenses/by/3.0/).

Contents lists available at ScienceDirect

International Journal of Pharmaceutics

journal homepage: www.elsev ier .com/locate / i jpharm

1. Introduction

According to recent WHO (world health organization) data(WHO, 2012a), 6.9 million of children under the age of five died in2011. The risk of dying for a child is highest in the neonatal period,and about 16.5 times higher for children under five in sub-SaharanAfrica compared to children in developed regions. WHO estimatesthat more than half of these early child deaths are caused byconditions that could be prevented or treated with existing simple,affordable interventions.

Pneumonia and malaria together cause the majority of deathsfrom infectious diseases, representing respectively 17% and 7% of

* Corresponding author at: University of Bordeaux, Faculty of Pharmacy, 146 rueLeo Saignat, Bordeaux 33076, France. Tel.: +33 557571229/224; fax: +33 557571197.

E-mail addresses: tina.kauss@u-bordeaux.fr, tikauss@yahoo.fr (T. Kauss).

http://dx.doi.org/10.1016/j.ijpharm.2014.04.0230378-5173/ã 2014 The Authors. Published by Elsevier B.V. This is an open access artic

deaths of children under five. Malaria deaths are caused byPlasmodium falciparum, and two-thirds of the pneumonia deathsare caused by Streptococcus pneumoniae; as a result these twoorganisms kill almost two million children each year in tropicalcountries (WHO, 2012a).

The difficulty in distinguishing severe malaria with acidoticbreathing from pneumonia in children has been well documented(Bergman et al., 2004; Hoban et al., 2001; O’Dempsey et al., 1993,1994). In general malaria is overdiagnosed and severe bacterialinfections are underdiagnosed. Clinical and pathology studiespoint to the diagnostic uncertainty and considerable overlapbetween severe malaria and sepsis. In up to 20% of children dyingwith an in-hospital diagnosis of cerebral malaria, a different causeis found at autopsy (Taylor et al., 2004) and both conditions coexistin 15–20% of cases (Berkley et al., 2005).

Therefore correct diagnosis and treatment cannot be expectedin villages and rural health centres where most cases occur. The

le under the CC BY license (http://creativecommons.org/licenses/by/3.0/).

Table 1Comparison of pH mobile phases composed by 80% of CH3OH and 20% of phosphatebuffer (30 mM).

wwpHof buffer s

wpHmobilephasea

kAZ kAM Asymmetry of AZpeak

RsAM/AZ

8.4 10.3 4.48 3.15 1.13 2.278.0 10.2 4.34 3.12 1.11 2.767.5 10.0 4.21 3.14 1.12 2.497.0 9.4 4.01 3.19 1.10 1.67

a Measurement performed at room temperature (i.e. 22 �C, air conditioned).

Table 2S/N ratio of compounds at various wavelengths.

Compound Wavelength (nm)

210 212 215

S/N

DHA 500 mg L�1 72 92 100AM 350 mg L�1 28 31 30AZ 500 mg L�1 220 201 172

56 A. Gaubert et al. / International Journal of Pharmaceutics 468 (2014) 55–63

integrated management of childhood illnesses guidelines, createdin 1992 by UNICEF and WHO, recommend presumptive malariatreatment for all children with fever higher than 37.5 �C in malaria-endemic areas and that both antibiotics and antimalarial drugs aregiven to seriously ill febrile children as pre-referral treatment ofmalaria or bacterial sepsis.

To prevent death, treatment must be available near home,outside hospitals, and as close to the village or household aspossible. Malaria and other febrile illnesses are frequently treatedat home in malaria endemic areas (Dolecek et al., 2008;Greenwood et al., 2007). When oral treatment is no longerpossible, and where injectable medications cannot be safelyadministered, rectal formulations can be administered by unqual-ified persons like parents or health volunteers. While rectalartesunate has been shown to be an effective pre-referraltreatment for malaria (Gomes et al., 2009), were both malariaand pneumonia are prevalent, a combined antibiotic–antimalarialtreatment is necessary. Treating malaria only might delay specifictreatments of other febrile diseases with similar symptoms, suchas sepsis and pneumonia (Whitty et al., 1999). Currently, systemiclarge spectrum antibiotics are available as oral or injectableformulations – of which none is suitable for pre-referral treatmentin pneumonia and malaria endemic regions. A fixed-doseantimalarial–antibiotic coformulation would be highly desirableand practical and was thus investigated.

The minimal characteristics of the target product profile (TPP)were: (i) active principles: well-known, in-use; physico-chemicalcompatibility between active ingredients when co-formulated; (ii)formulation: uncomplicated, easy to scale-up, inexpensive tomanufacture; well-known, inexpensive excipients; rapidly bio-available; (iii) product development: simple, rapid, inexpensive;(iv) stability: suited to tropical conditions; (v) price: low-cost (bothcost-of-goods and final product); (vi) efficacy: rapidly actingantimalarial and broad-spectrum antibiotic (to include mainbacterial species causing sepsis and pneumonia in neonates,infants, toddlers and children), with as little resistance existing aspossible; (vi) safety: proven general safety profile of activeingredients. The preliminary, pre-clinical development of suchformulation was described in this article.

2. Material and methods

2.1. Material

Artesunate (AS), AM, AZ and dihydroartemisinin (DHA) werepurchased from Knoll BASF Pharma (Liestal, Switzerland), Sanofi(France), Pfizer (USA) and Sigma–Aldrich (Saint-Quentin Fallavier,France), respectively. Excipients were of pharmaceutical grade.Sodium laurylsulfate (SLS), colloïdal silica (Aerosil 300), sodiumcroscarmellose, polyvinylpyrrolidone (PVP), talc and magnesiumstearate were pourchased from Cooper (France). Different grades ofPEG (polyethylenglycol) were purchased form Fagron (France).Microcrystaline cellulose (Avicel PH302) was purchased from FMCBiopolymer (Ireland). Betain and Lutrol were purchased form VWR(france) and BASF (Germany) respectively. Analytical solvents,acetonitrile and methanol were isocratic HPLC grade, purchasedfrom Prolabo VWR (Leuden, Belgium). KH2PO4 and Na2HPO4, 12H2Owere from Merck (Darmstadt, Germany).

2.2. HPLC analysis and method optimization

2.2.1. HPLC conditionsThe liquid chromatography system consisted of a Spectra

System P4000 pump, a UV 6000LP detector with a cell path lengthof 50 mm, an AS 3000 autosampler and a SN 4000 systemcontroller from TSP (Courtaboeuf, France). Column was Luna C8 (2)

100 A EC 5 mm, 150 � 4.6 mm (Phenomenex, France) thermostatedwith 560-CIL (Cluzeau Info Labo, Saint Foy la Grande, France). Theflow rate was 1 mL min�1. The sample injection volume was 5 mL.

For method optimization, several mobile phases and detectionwavelenghts were screened (Tables 1 and 2).

2.2.2. Solubility study of AM and AZ in presence of lutrolIn a dissolution tester three bowls were filled with 1 L of

phosphate buffer 15 mM pH 8 to simulate rectal pH conditions.Accurately weighed amounts of approximately 400 and 300 mg ofAZ and AM were respectively added to each vessel. Lutrol wasadded to two of the three vessels at 2% and 5% of tablet mass (1.4 g).Samples were taken at time zero and at time points of 1 h, 2 h and4 h. Homogeneity of solutions was tested by sampling each timepoint twice.

Determination of active pharmaceutical ingredient (API)content was obtained by calibration curves of AM and AZ at fivedifferent concentration levels (10%, 40%, 70%, 100% and 120%) inpresence or absence of lutrol at 5%. Stock solution for eachconcentration level of AZ and AM were prepared by dissolving anaccurately weighed amount of each drug and/or excipient in 50 mLof mobile phase then filtrated in a 0.45 mm filter. One milliliter ofthe filtrated solution of each concentration level was diluted withphosphate buffer pH 8 up to 20 mL. A 100% standard solutioncontained 400 and 300 mg L�1 of AZ and AM, respectively. Thesevalues were based on the theoretical tablet drug content. Allcalibration curves were proven to be linear. Furthermore, nodifference was observed in presence or in absence of lutrol at 5%.

2.3. Formulations tested

2.3.1. Compatibility study between AZ and AMCompatibility study between both APIs, namely AZ and AM, was

performed. Binary physical mixture of precisely weighted APIs wasmixed using Turbula (France) at 67 rpm for 10 min. The mixturewas divided into three parts, T0, ambient condition and accelerated(40 �C) condition. Samples were taken up till 3 months andanalyzed using HPLC method as described.

AM and AZ compatibility was further evaluated usingdifferential scanning calorimetry (DSC). DSC analysis was per-formed using Mettler Toledo TA controller and DSC30,(Switzerland) with STARe software. DSC method consisted in aheating rate of 5 �C min�1 in the range of 30–180 �C. Samples of 6–8 mg were precisely weighted in aluminium pans, sealed and

A. Gaubert et al. / International Journal of Pharmaceutics 468 (2014) 55–63 57

perforated with a pin. An empty sealed perforated aluminium panwas used as reference.

2.3.2. Compatibility study between APIs and excipientsExcipients chosen were checked for their usage among the ones

already used in commercial formulations with AZ and AM.Compatibility was considered acquiered for excipients commer-cially available with both APIs separately. In addition, Am and AZcompatibility study with lutrol and talc were carried out. Sixsamples of AM, AZ or both APIs were added to lutrol or talcseparately, then mixed (Turbula, France) at 67 rpp for 10 min andkept at 40 �C during 3 months. Samples were taken at time zeroand then at 1 month, 2 months and 3 months. AM and AZ at threedifferent concentrations levels (80%, 100% and 120%) wereprepared in presence and in absence of excipient tested. Stocksolutions for each concentration level of AZ and AM were preparedby dissolving an accurately weighed amount of each drug and/orexcipient in 50 mL of methanol then filtrated with 0.45 mm nylonfilters. Five milliliters of the filtrated solution was diluted up to10 mL with the HPLC mobile phase. A 100% standard solutioncontained 500 mg L�1 of AZ and AM, each. All calibration curveswere linear. Calibration curves statistically not different inpresence or in absence of lutrol and talc.

2.3.3. Preparation of tested formulationsAll tested formulations (Table 3) were prepared in our

laboratory. ASAM gels were prepared either dispersing gellingexcipient in the aqueous solution containing both APIs oralternatively, by preparing a blank gel and then adding successivelyboth APIs. Gel liquefaction occurred rapidly in both cases.

PEG suppositories were prepared by dispersing both APIs in themelted PEG under stirring. Formulation was molded in suppository

Table 3Summary of tested formulations for rectal administration of antimalarial–antibiotic co

APIs Formulation Immediatecompatibility

Stability Animpharkine

Artesunate + azithromycin Gel (severalformulations)

No No ND

PEG suppo-sitory

PEG 6000monolayer orbilayer

No No

Eudragit1 encapsulatedAS could beconsideredfor PEGinclusion,howeverproblems offormulationencapsulateddrug capacity

PEG1500/4000

Yes Interrupted(colourchange)

Hard capsule Yes No ND No gArtemether + azithromycin PEG suppository No ND NDb

Hard capsule With LSa No No

LSa–AMinteractionDouble

granulationnecessary

WithoutLSa

Yes Yes Yes

Rectodispersible tablet Yes Yes Low AMabsorption

No g

ND = not determined.a LS = Na laurylsulfate.b AM suppository in monotherapy gave no absorption.

moulds, left at room temperature until solidification, and thenleveled before taking them out of the moulds. Suppositories werestored in a desiccator.

Hard gelatin capsules (HGC) were prepared by filling thepowder mixture containing both APIs, a filler and a lubricant(mixed using Turbula at 67 rpm for 10 min) into gelatin capsules(size 000), using a semiautomatic capsule filler (LGA, QB300,France).

Rectodispersible tablets were prepared by direct compressionusing laboratory scale alternative compressor (Korsch pressen,type EK0, Berlin, Germany).

2.4. In vitro evaluation of formulations

Common pharmacotechnical tests (uniformity of mass, tablethardness and friability, etc) were performed following Europeanpharmacopoeia 8.0 methods and criteria.

2.4.1. Dissolution assay for AZ and AM cotherapy dosage formsThe apparatus used was paddle Sotax AT 7 (Basel, Switzerland)

at 37 �C � 0.5 �C and 50 rpm. Various media were tested (cf. resultSection 3.3). Finally, a bi-phasic dissolution method, initiallydeveloped for AM tablets (Gabriëls and Plaizier-Vercammen,2004), was adapted for AMAZ cotherapy. Briefly, 150 mL of50 mM phosphate buffer pH 6.0 and 100 mL of isooctane wereplaced into teach dissolution bowel. Preliminary data showed thatthe pH variation between 5.5 and 7.0 did not modify AZ and AMdissolution profiles significantly. Phase separation was at thepaddle level, ensuring the agitation of both phases. Each form(ballasted capsule or tablet, n = 6) was placed at the bowel bottom(into the aqueous phase). At defined times (0/10/20/30/45/75 min),samples of 0.5 mL buffer and 0.5 mL isooctane media were taken

therapy and conclusions.

almaco-tics

State Remarks

No go Gel liquefaction; could propose extemporaneous preparationtype DIASTAT1, but high additional costs

ND Stopped

o AS degradation after 6 weeks at 40�C/75% RHStoppedb Problems of drug release consecutive to AM-PEG interaction

(shell formation); 0.1% of LSa in dissolution medium or 1% LSa

in suppository are necessary for in vitro drug release.ND No go

Potentialcandidate

o Low absorption may be caused by aggregation between AM andAZ.

58 A. Gaubert et al. / International Journal of Pharmaceutics 468 (2014) 55–63

simultaneously at the interface using 10 mm PTFE pre-filterequipped syringes. 300 mL of organic phase was evaporated at60 �C and re-dissolved in 1550 mL of HPLC mobile phase, thencompleted with 450 mL of buffer phaseand mixed before injectionin HLPC system.

2.4.2. Preliminary stability studyPreliminary stability study was performed on several formu-

lations: ASAZ capsules, AMAZ tablets and capsules. Preparedbatches of each formulation were divided into T0 condition,ambient condition and accelerated aging condition (40 �C, 75% RH)using in a climatic chamber (Froilabo/Frilabo, France). All formswere placed into Bakelite stoppered glass vials containing each 30dosage forms (5 vials per condition). At defined time (0/1/2/3/6months for capsules and 0/1/2/3 months for tablets), vials weretaken for aspect, content and drug release analysis (0 and 6months). For AM and AZ content determination, crushed tablets(n = 6) or capsule content (n = 10) was dissolved directly in HPLCmobile phase, then analyzed using HPLC method described.

2.5. In vivo bioavailability evaluation

2.5.1. Animal experimentsAnimal experiments were performed at Etablissement Français

du Sang (EFS, Aquitaine-Limousin, Bordeaux, France; accreditationnumber for animal experimentation n�A33063080).

Five adult healthy New-Zealand white rabbits, provided by « Legrand Claud », (Eyvirat, Dordogne, France), were used for thepharmacokinetic study of each form. Animals were housed inindividual cages in controlled temperature (18–21 �C) and humidi-ty (40–70%) room. There were fastened for 24 h prior and duringthe experimentation but allowed free access to water and glucosesolution (5%).

The administration of each pharmaceutical formulation wasadapted individually to the rabbit body weight at 20 mg/kg of AMand AZ. AM and AZ (monotherapy) in miglyol were used as rectalcontrol formulations.

The protocol complied with the European Community guide-lines (EU directive 2010/63/EU for animal experiments) asaccepted principles for the use of experimental animals andinternal ethic principles.

Blood samples from peripheral ear vessels (at least 1 mL ofblood per time point) were collected into heparinized plastic tubesby inserting a 22 GA I.V. catheters (BD Insyte1, Spain). They werecollected before administration, then 12 samples in 48 h post-administration. Blood samples were kept on ice before centrifuga-tion at 1200 g during 10 min within 30 min after collection. Onlysupernatants (at least 500 mL) were transferred into cryo tube andconserved at �80 �C prior analysis.

2.5.2. PK sample analysisSamples were analyzed using previously described methods for

AM and AZ (Kauss et al., 2012; Tarning et al., 2012).

2.5.3. PK data processingThe maximal concentration (Cmax), and the time to reach the

maximal concentration (Tmax) of each formulation were takendirectly from mean plasma concentration–time profile curves,whereas the area under the concentration–time curve (AUC) wascalculated using trapezoidal rule. Standard deviations are given forall PK parameters. Relative bioavailability was estimated frommean AUC0–48 h. Data at 24 h were analyzed by comparison withpreviously obtained parameters (Kauss et al., 2012).

2.6. Powder X-ray diffraction

Crystallinity and phase identification were obtained by powderX-ray diffraction analysis (PXRD) performed on a D5005 Bruker/Siemens diffractometer using the theta/theta geometry withcopper radiation. Samples were deposited on a stainless steelholder without grinding to avoid artificial compression effects.Data were collected for 2u angle range of 3–40� with a step size of0.02� at a scanning speed 0.06�/min. Mixture samples were thengrinded in order to compare the grinding effect with thecompression in tablets products.

2.7. Statistical analysis

Data were analyzed using Student bilateral paired T test. Thedifference of p < 0.05 was considered significant.

3. Results and discussion

3.1. TPP and the choice of APIs

Considering the abovementioned TPP, only drugs alreadyregistered and in-use were taken into account.

For malaria, artemisinin derivatives are the most potent andrapidly acting antimalarial drugs available; they are very safe andwell tolerated, and there is no significant resistance yet outsideareas of SE Asia (WHO, 2012b). They are the WHO-recommendedfirst-line treatment of uncomplicated malaria as artemisinin basedcombination treatment, (ACTs), severe malaria and pre-referraltreatment (WHO, 2010). Based on generally available informationand significant experience accumulated in our laboratory onphysico-chemical properties and stability of artemisinin deriva-tives, the choice was between AM and AS; AM was expected tohave stability advantages, against the considerable experience andtrack record existing already on AS.

Furthermore, protecting the artemisinin derivatives againstresistance when deployed is also a major concern, and ideally thecompanion antibiotic should also have antimalarial properties.These principles informed the choice of the antibiotic. Ceftriaxone(CFX) would be the lead candidate based on antibacterialcharacteristics, but has minimal pharmaceutical information(which meant a longer pharmaceutical development time), hada relatively high cost-of-goods, and importantly no antimalarialactivity. Of the other two contenders initially considered,chloramphenicol was dropped for clinical (real and perceivedsafety issues; spectrum; resistance) and compatibility reasons(incompatible with AS in preliminary studies). Eventually, AZ wasselected for further investigation as the antibiotic partner in thefixed-dose antimalarial–antibiotic combination, for its expandedantimicrobial spectrum, covering most bacteria responsible for themain causes of children mortality and including malaria’s parasitePlasmodium falciparum, for its improved pharmacokinetic charac-teristics over older macrolides, (Andersen et al., 1995; Dunn andBarradell,1996; Langtry and Balfour, 1998; Noedl et al., 2007, 2001;Yeo and Rieckmann, 1995) and for the considerable experiencegained on AZ rectal forms (Kauss et al., 2013, 2012).

3.2. Analytical control method development

3.2.1. Combined dosage of AS and AZSeveral methods were previously developed to test various

combined ASAZ formulations. Some of them were published, likegel investigation (Gaudin et al., 2008) and HGC investigation(Boyer et al., 2012) using HPLC/diode array detector (DAD), HPLC/evaporative light scattering detection (ELSD) or near infraredspectroscopy. For PEG suppository and Eudragit1 encapsulation

Table 4Final formulations for of AMAZ capsules and tablets.

API/excipient (% w/w) AMAZ HGC AMAZ tablet

Azithromycin 2H2O(anhydrous)

51.05(48.74)

29.13(27.79)

Artemether 36.56 20.87Microcrystalline cellulose 12.19 41.3Croscarmellose sodium – 4.0Lutrol F68 – 2.0Colloidal silica 0.20 0.2Talc – 2.0Mg stearate – 0.5

A. Gaubert et al. / International Journal of Pharmaceutics 468 (2014) 55–63 59

investigation the HPLC/DAD was found to be specific and thus usedfor control quality.

3.2.2. Combined dosage of AM and AZAs part of previous work, a method had been developed for the

simultaneous content determination of AZ and AM in a supposi-tory (Gaudin et al., 2011). Beside AM and AZ, DHA, the activemetabolite and main degradation product of artemisinin deriva-tives, needed to be dosed. Its presence was expected in lowconcentrations; therefore the sensitivity of its detection wasstudied.

Mobile phase was optimized in order to improve the columnlife-span. The column was a grafted silica resistant to basic pH, aLuna C8(2) which was stable in the range of s

wpHw from 1 to 10.

Decreasing wwpH

w from 8.4 to 7.0 at 25 �C led to a decrease ofapparent pH of the mobile phase, AZ retention and AM–AZresolution whereas AZ peak asymmetry was not significantlydifferent (Table 1). In order to keep sufficient resolution betweenAZ and AM a value of w

wpHw equal to 7.5 was selected. The column

was then thermostated at 20 �C in order to improve methodrobustness. In this condition, DHA was eluted as a unique peak,improving its sensitivity, with a retention factor equal to 1.37.Table 2 shows the signal-to-noise (S/N) ratio for AM, AZ and DHAobtained at various wavelengths. S/N ratio of AM was almostconstant while it decreased from 210 to 215 nm for AZ. DHA S/Nratio varied in the opposite direction, with a maximum at 215 nm.Therefore, 215 nm was selected as the optimal wavelength for thesimultaneous detection of AM, AZ and DHA.

The lower limit of quantification (LOQ) for AM, AZ and DHAwere 25, 25 and 50 mg L�1, respectively.

Method validation was then performed for rectodispersible andHGC formulations and was used for the solubility, compatibilityand stability studies.

3.3. Preformulation of antimalarial–antibiotic cotherapy

Artemisinin derivatives, AS or AM, were tested for theircompatibility with AZ in a dry powder mixture at 40 �C and werefound compatible (HPLC analysis, DSC) after 6 months. Allexcipients were either already commercialized in AZ and AM orAS formulation and therefore considered as compatible, or testedfor their compatibility in physical mixtures to support their choicefor coformulations. As the stress degradation study of theanalytical method validation, performed on APIs, had showedthat AZ and AM are unstable in acidic conditions, excipients withacidic functions were discarded. The formulations retainedfollowing preliminary explorations are summarized in Table 3.Their compatibility/feasibility, preliminary stability at 40 �C/75%RH and, for selected ones, bioavailability were evaluated.

An important point of in vitro evaluation was in vitro drugrelease assay. AZ aqueous solubility (39 g L�1 at pH 7.4 and 37 �C(Pfizer, 2013)) was not limiting regarding the dose considered(421 mg AZ dihydrate). In contrast, artemisinin derivatives arecharacterized by their low aqueous solubility: 0.296 g L�1 (Kausset al., 2010) and 0.161 g L�1 (Gabriëls & Plaizier-Vercammen, 2004)for AS and AM respectively at 37 �C. For dosage forms containing300 mg of an artemisinin derivative, this solubility was insufficientto meet solubility (ideally sink, i.e. 3.0 g L�1, but minimally0.90 g L�1) conditions of the dissolution assay. The dissolutiontest was developed targeting AM, since it has lower aqueoussolubility than AS. Varying the buffer pH (5.5–7.5) did not result ina satisfactory increase in AM solubility. The addition of 1% w/v ofSLS to the buffer at pH 7 was necessary to obtain requested AMsolubility, but this changed the dissolution rate (drug release wasslower for both AM and AZ) as a consequence of the increasedviscosity of the medium. The Gabriëls and Plaizier-Vercammen

(2004) octanol–buffer biphasic system was adopted as the bestcompromise between physiological conditions and pharmacopoe-ia specifications. Of note, as this method required that each samplewas treated individually, the variance of results was inherentlyhigher than in a conventional monophasic dissolution assay.

The solubility of each API in the other phase i.e. AM in the bufferphase and AZ in the isooctane phase was showed to be low(respectively 1.1 �0.2% and 1.4 � 0.1% of considered dose). Bothphases were treated separately but reconstituted in a single samplebefore analysis, to give the total amount of both drugs.

3.4. Formulation and in vitro characterization of antimalarial–antibiotic coformulation

Several pharmaceutical forms were considered: classicalsuppository form, dry forms like capsules and tablets and semi-solid forms, like rectal gels. We have already successfullydeveloped an AZ rectal formulation (Kauss et al., 2013, 2012).The additional challenge for developing a coformulation consistedin obtaining simultaneously good stability and bioavailability ofAPIs (AZ and AS or AM) with different physicochemical profiles.

ASAZ and AMAZ feasibility studies (Table 3) showed lack of ASstability. ASAZ gels suffered from liquefaction as soon as both APIswere included. ASAZ PEG suppositories presented a colour changeand the stability study was interrupted. AS was encapsulated inseveral Eudragit1 polymers for protection against degradation;however, the high apparent density of encapsulated AS renderedits inclusion in suppositories and HGC technically difficult. As thecompatibility of a dry blend was showed, an extemporaneouslyreconstituted gel was considered. Such formulation required aparticular device for its reconstitution/administration. A conve-nient, commercially available device was the one used fordiazepam rectal gel (DIASTAT1 AcuDialTM of Valeant Pharmaceut-icals North America), but the additional cost of this primaryconditioning did not comply with our TPP. Accordingly, AS basedcoformulations were abandoned for stability (gels, suppositories,HGC) and/or cost (extemporaneously reconstituted formulations)issues in favour of AM based coformulations.

Dry AMAZ rectal formulations were selected for furtheroptimization. First, a simple AMAZ dry powder mixture wasconsidered to be filled in HGC. Commonly, HGC are used mainly byoral route, but could be used for rectal route as well. Two aspectswere considered of particular import during optimization: thesolubility of AM to be enhanced by adding surfactants, and theflowability of the powder mixture to facilitate the industrial scaleup.

Several AMAZ hard capsule formulations were developed usinglactose as filler, with or without SLS, or PVP. Because of HGCcapacity, and in order to achieve appropriate flowability, theseformulations needed to be granulated or double-granulated(necessary in case of SLS) before filling in gelatine cores.Surfactants containing formulations gave erratic, non-reproduc-ible drug content and drug release results. These results were

Table 5AM compatibility with surfactants after 3 months at 40 �C (HPLC analysis of AMdrug content).

Mean AM content (%) Coefficient of variation

AM + SLS 101.14 10.83AM + lutrol F68 96.06 5.40AM + betain 83.86 0.14AM + PEG 6000 102.61 2.13

60 A. Gaubert et al. / International Journal of Pharmaceutics 468 (2014) 55–63

attributed partially to the double-granulation process and to apossible interaction between AM and SLS. Finally, surfactants wereexcluded from HGC formulation and microcrystaline cellulose wasused as flowability-promoting filler in addition to a lubricant.Among common lubricants, colloidal silica gave the best resultscompared to talc and magnesium stearate (see Table 4 for finalformulation).

Beside the HGC coformulation, a rectodispersible tablet wasconsidered. To optimize the bioavailability of both APIs, theunderlying rationale of this approach was to (i) avoid the shell (i.e.capsule) and consequently its disintegration delay, (ii) ensure arapid disintegration in an environment (the rectum) with littleamounts of liquids, and (iii) enhance the solubility and conse-quently the absorption of AM by adding a surfactant. Whenpossible, the excipients choice was made among those used inexisting commercial formulations to avoid compatibility andregulatory issues. The key points of the development were tablets’characteristics (e.g. hardness, friability . . . ), rapid disintegrationand AM solubility enhancement. For the latter, surfactantcompatibility studies were performed (cf. Table 5). PEG 6000was excluded due to the suspected interaction of AM with PEG1500 and 4000 in preliminary suppository formulation studies(Table 3) preventing efficient drug release. SLS gave highly variableresults and Betain was associated with AM drug content decrease.Lutrol 68 was selected as surfactant for further studies ofrectodispersible AMAZ tablet and no feasibility problems werenoticed. Its use was validated for the optimized formulation ofAMAZ rectodispersible tablets.

The characteristics of the final formulations retained for furtherin vitro evaluations are summarized in Table 5.

AMAZ tablets and capsules were evaluated for customarypharmacotechnical controls (mass and content uniformity, tabletfriability, disintegration, etc.) according to the European Pharma-copoeia 8.0 requirements. As there is currently no specificmonograph for rectodispersible tablets, tablet disintegration wastested following the monographs of dispersible and vaginal tablets.All the tested parameters complied with pharmacopoeia stand-ards.

Dissolution assay and preliminary stability studies were alsoperformed on these formulations, up to 3 or 6 months for tabletsand HGC respectively (Table 6).

The preliminary stability results showed that contrary toASAZ formulations (degradation of AS after 6 weeks), AMAZformulations were stable for at least 3 months in acceleratedconditions.

Table 6In vitro evaluation of AMAZ formulations (n = 6, mean � SD).

Time (weeks) Storage condition

Drug content (%) T0 –

12 ambient

12 40�/75% RH

26 40�/75% RH

Drug released at 45 min(%) T0 –

26 40�/75% RH

ND: not determined.

The drug release of both APIs from HGC and AZ from tabletscomplied with pharmacopoeia specification of conventional drugrelease, while drug release was low for AM from rectodispersibletablets – a surprising finding, given the addition of a super-disintegrant and a surfactant to the formulation. As the in vitrobiphasic assay was not representative of physiological conditionsregarding rectal mechanical forces, the formulation was tested invivo despite in vitro results. Drug release of both API from HGC wasnot significantly different (p > 0.05, Student test) after 26 weeks ofaccelerated stability compared to the initial release.

3.5. In vivo evaluation

AMAZ coformulations were administered rectally to rabbits andtheir bioavailability evaluated. AZ suspension and AM solution inoily vehicles were used as rectal controls. The results aresummarized in Fig. 1A and B for AZ and AM plasma profilesrespectively and in Table 7 for pharmacokinetic parameters.

For ethical reasons this preliminary bioavailability study waslimited to groups of five animals per experimental condition,which showed marked variability. Further animal studies con-taining larger groups of animals would be necessary to reduce thevariability for the selected formulation, but this was beyond thescope of the current preliminary exploratory study.

AMAZ HGC formulation produced satisfactory absorption for AZ(comparable AUC and relative bioavailability F0) and delayed butenhanced absorption of AM (relative bioavailability of 146%,difference not statistically significant due to large standarddeviation) compared to rectal single-agent control formulations.Compared to previously described AZ monotherapies (Kauss et al.,2013, 2012), HGC parameters were comparable (107% and 105% ofrelative bioavailability for AZ HGC in monotherapy and cotherapyrespectively), but AZ HGC absorption was lower than the oneobtained with AZ suppository (232% of relative bioavailability after24 h) (Kauss et al., 2012). In conclusion, AZAM HGC provided AZabsorption comparable to the one obtained with rectal oilysuspension, but did not enhance it. AM alone in HGC was alsoadministered to rabbits in a monotherapy as a rectal dry formcontrol condition and gave statistically non different results (AUC187 � 71 ng � h/mL, p > 0.05) as AM absorbed from cotherapy.Furthermore, these results demonstrated that an additional API didnot diminish the absorption rate of AM or AZ form HGC.

AMAZ tablet moderately but non-significantly enhanced AZabsorption (relative bioavailability 133%, p > 0.05), but failed toprovide sufficient AM absorption (relative bioavailability 25%,p � 0.05). These results confirmed the ones obtained during in vitroevaluation. Despite the presence of a disintegrant and a surfactant,AM was not adequately absorbed in rabbits.

3.6. Further investigations to explain low AM release of AMAZ tablets

Unexpectedly low AM drug release from AMAZ tablets wasfurther investigated. Possible explanations are that (i) crystallinemodification of AM in tablet form diminished its drug release and

AMAZ HGC AMAZ tablet

AZ AM AZ AM

100.4 � 1.3 101.2 � 1.1 98.7 � 1.6 97.6 � 1.1101.7 � 1.0 100.0 � 0.9 101.2 � 0.8 100.3 � 2.7100.8 � 0.6 99.0 � 0.2 99.2 � 1.8 98.3 � 2.797.3 � 0.8 100.9 � 1.5 ND ND88.0 � 15.0 78.9 � 11.0 22.6 � 3.3 88.8 � 0.893.4 � 3.6 87.1 � 11.4 ND ND

Fig. 1. AZ (A) and AM (B) rabbit plasma profiles of various rectal formulations administered at 20 mg/kg body weight.

Table 7AZ and AM disposition in the rabbit after administration of individual drugs asmonotherapy as control (AZ or AM in miglyol) or fixed-dose coformulations (n = 5,mean � SD).

Control AMAZ HGC AMAZ tablet

AM parametersCmax (ng/mL) 35 � 11 56 � 18 59 � 38Tmax (h) 1.4 � 0.6 3.0 � 1.2 0.8 � 0.2AUC 0–48 h (ng � h/mL) 126 � 85 185 � 130 46 � 19F0 100 146 25

AZ parametersCmax (ng/mL) 191 � 49 118 � 35 410 � 212Tmax (h) 0.37 � 0.32 1.25 � 0.89 0.80 � 0.19AUC 0–48 h (ng � h/mL) 858 � 198 898 � 469 1141 � 404F0 100 105 133

A. Gaubert et al. / International Journal of Pharmaceutics 468 (2014) 55–63 61

its bioavailability and (ii) lutrol added as a surfactant did notsignificantly increase AM solubility – at least not enough tocompensate the decrease induced by crystalline modification.

3.6.1. X-ray characterizationPowder X-ray diffraction (PXRD) patterns were determined on

powder of AZ, AM, AMAZ mixture and tablets of AZ, AM and AMAZmixture. Each PXRD pattern presented the characteristic peaks ofAZ or AM or both depending on the sample, and no spurious extrapeaks.

All PXRD patterns of samples containing AM, except the AMAZtablet, showed a strong preferred orientation effect of the AMphase as shown by the abnormally intense 020 reflection of AM(Fig. 2A), which may be due to strong anisotropy of the shape of theAM crystallites. It is worth to note that this effect, which decreasedmarkedly in the compressed mixture of AMAZ tablets, was presentin all other samples, including the compressed AM (Fig. 2B). Thisdifference, between the PXRD pattern of the compressed mixturetablets and the others, constituted the sole distinct feature withinthe observed patterns.

In order to better understand this effect, a Rietveld refinement(Rietveld, 1969) was performed on each PXRD pattern. The

refinement performed on the AMAZ mixture samples led to thesame weight proportions of AM and AZ in the powder and thetablet samples which were estimated at 40(2)% and 60(2)%respectively – thus in good agreement with the theoreticalproportions of 42% and 58%, respectively. These results allowed

Fig. 2. Powder X-ray diffraction patterns of AM, AZ and AM/AZ mixture of (A) powder samples and (B) compressed tablet samples.

62 A. Gaubert et al. / International Journal of Pharmaceutics 468 (2014) 55–63

to exclude any chemical interaction or decomposition of AM or AZand any amorphization during the compression.

The Rietveld refinement of pure compounds confirmed thestrong effect of preferred orientations on both powder and tabletAM samples, while no significant orientation effect could bedetected on AZ samples. The preferred orientations effect were ofthe same amplitude between AM powder and tablet samples(March-Dollase parameter �0.6) (Dollase, 1986), but interestingly,the Rietveld refinement on AMAZ mixture revealed a significantlydifferent behavior: while the preferred orientations effect was stillstrong in the powder mixture (March-Dollase parameter �0.5),

Fig. 3. Dissolution study in the presence of lutrol for AZ (A) and AM (B).

this effect completely disappeared in the tablet mixture (March-Dollase parameter �1.0). This indicated that the loss of thepreferred orientations was only due to the compression of thepowder mixture. According to the Rietveld results the distinctbehavior of the AMAZ mixture tablets relative to the powdermixture might be related to the sole preferred orientations effectand cannot be attributed to changes in the AM and AZ proportionsin the samples. The total loss of the preferred orientations effectafter compression of the mixture might be due to a change ofcrystallites morphology, which was possibly due to either thebreaking of the crystallites into smaller pieces or by theaggregation of crystallites of AM and AZ. In either case the shapeof the new particles became more isotropic, thus reducing thepreferred orientations. Nevertheless, the first hypothesis wascontradicted by the absence of this effect in the pure AM tabletsamples and by low absorption of AM tablets on pharmacokineticmeasurements. Thus aggregation was a more likely explanation.This hypothesis was further supported by finding in the Rietveldrefinements of a small broadening of the diffraction peaks at hightheta value for the AMAZ mixture tablet sample only, which wasconsistent with micro-deformations of the material caused byaggregation of AM and AZ crystallites. As grinding of the AMAZpowder mixture did not lead to the suppression of preferredorientations, this aggregation would occur in tablet samples only,and might be due to local surface warming during compression.

Taken together, these results demonstrate that compressionhad an effect on the crystalline microstructure and could help toexplain the observed differences in AM release between com-pressed and uncompressed forms.

3.6.2. AM and AZ solubility enhancement in presence of lutrolTo better characterize the lutrol effect as a surfactant in

enhancing the AM and AZ solubility, the latter was studied in thepresence of 2% lutrol, at tablet dosage form quantities (300 and400 mg for AM and equivalent anhydrous AZ, respectively).

AZ dissolution rate improved with lutrol at 2% (57% vs. 42%respectively after 1 h, p < 0.05). After 4 h with 2% lutrol, 76% of theexpected concentration was reached while compared to 60%without lutrol (Fig. 3A). Conversely, 2% lutrol has no effect on AMsolubility or dissolution rate in tested conditions (Fig. 3B). Similarresults were found for the AMAZ tablet. Additional studiesdemonstrated that AM dissolution increased with higher concen-trations (5%) of lutrol (from 25% to 30% for 0% and 5% of lutrolrespectively, p < 0.05). Further studies are needed now on moreefficient surfactants to improve AM solubility and dissolution rateon order to compensate the decreased drug release induced bycompression.

A. Gaubert et al. / International Journal of Pharmaceutics 468 (2014) 55–63 63

4. Conclusions

Our study demonstrates that dry pharmaceutical forms are themost suited coformulation for an antimalarial–antibiotic combi-nation treatment in malaria-endemic areas. They provide an easyto administer, inexpensive, stable formulation, with enhanced AMrectal bioavailability and comparable AZ bioavailability withrespect to individual rectal preparations. The co-administrationof individually-formulated antimalarial and azithromycin (e.g. ASrectal soft capsule Rectocap1 and the previously developed AZsuppository (Kauss et al., 2013)) remains a valid, shorter-termalternative.

Rectodispersible AMAZ tablets gave satisfactory AZ but low AMdrug release and bioavailability, possibly due to compression-induced changes in AM crystalline micro-structure.

Acknowledgements

The authors wish to thank the members of Laboratoire deTechnologie Industrielle de Bordeaux, University of Bordeaux, DrsPascale Gueroult, Gilles Lemagnen and Mr Denis Larrouture andM2 students Marine Brepson, François Fagour and Sarah Noury, fortheir help for rectodispersible tablets development; and Dr ZoranIvanovic, Scientific director of EFS Aquitaine Limousin, France, forhis help regarding animal experimentation.

This work was supported in part by the Wellcome Trustfoundation, Wellcome Trust Feasibility Award ref: 085242/Z/08/Z.

P. Olliaro is a staff member of the WHO; the authors alone areresponsible for the views expressed in this publication and they donot necessarily represent the decisions, policy or views of the WHO.

References

Andersen, S.L., Ager, A., McGreevy, P., Schuster, B.G., Wesche, D., Kuschner, R., Ohrt,C., Ellis, W., Rossan, R., Berrman, J., 1995. Activity of azithromycin as a bloodschizonticide against rodent and human plasmodia in vivo. American Journal ofTropical Medicine and Hygiene 52, 159–161.

Bergman, M., Huikko, S., Pihlajamaki, M., Laippala, P., Palva, E., Huovinen, P., Seppälä,H., 2004. Finnish study group for antimicrobial resistance (FiRe Network). Effectof macrolide consumption on erythromycin resistance in Streptococcus pyogenesin Finland in 1997–2001. Clinical Infectious Diseases 38, 1251–1256.

Berkley, J.A., Lowe, B.S., Mwangi, I., Williams, T., Bauni, E., Mwarumba, S., Ngetsa, C.,Slack, M.P., Njenga, S., Hart, C.A., Maitland, K., English, M., Marsh, K., Scott, J.A.,2005. Bacteremia among children admitted to a rural hospital in Kenya. NewEngland Journal of Medicine 352, 39–47.

Boyer, C., Gaudin, K., Kauss, T., Gaubert, A., Boudis, A., Verschelden, J., Franc, M.,Roussille, J., Boucher, J., Olliaro, P., White, N.J., Millet, P., Dubost, J.-P., 2012.Development of NIRS method for quality control of drug combinationartesunate – azithromycin for the treatment of severe malaria. Journal ofPharmaceutical and Biomedical Analysis 67–68, 10–15.

Dollase, W.A., 1986. Correction of intensities for preferred orientation in powderdiffractometry: application of the March model. Journal of Applied Crystallog-raphy 19, 267–272.

Dolecek, C., Tran, T.P., Nguyen, N.R., Le, T.P., Ha, V., Phung, Q.T., Doan, C.D., Nguyen, T.B., Duong, T.L., Luong, B.H., Nguyen, T.B., Nguyen, T.A., Pham, N.D., Mai, N.L.,Phan, V.B., Vo, A.H., Nguyen, V.M., Tran, T.T., Tran, T.C., Schultsz, C., Dunstan, S.J.,K, Stepniewska, Campbell, J.I., To, S.D., Basnyat, B., Nguyen, V.V., Nguyen, V.S.,Nguyen, T.C., Tran, T.H., Farrar, J., 2008. A multi-center randomised controlledtrial of gatifloxacin versus azithromycin for the treatment of uncomplicatedtyphoid fever in children and adults in Vietnam. Public Library of Sciences 3,e2188.

Dunn, C., Barradell, L.B., 1996. Azithromycin – a review of its pharmacologicalproperties and use as a 3-day therapy in respiratory tract infections. Drugs 51,483–505.

Gabriëls, M., Plaizier-Vercammen, J., 2004. Design of a dissolution system for theevaluation of the release rate characteristics of artemether and dihydroarte-misinin from tablets. International Journal of Pharmaceutics 274, 245–260.

Gaudin, K., Kauss, T., Gaubert, A., Viaud, V., Olliaro, P., White, N.J., 2011. Simultaneousdetermination of artemether and azithromycin in suppository by reversedphase HPLC. Analytical Letters 44, 2732–2743.

Gaudin, K., Barbaud, A., Boyer, C., Langlois, M.-H., Lagueny, A.-M., Dubost, J.-P.,Millet, P., Fawaz, F., 2008. In vitro release and stability of an artesunate rectal gelsuitable for pediatric use. International Journal of Pharmaceutics 353, 1–7.

Greenwood, D., Finch, R., Davey, P., Wilcox, M., 2007. Antimicrobial Chemotherapy,fifth ed. Oxford University Press, Oxford.

Gomes, M.F., Faiz, M.A., Gyapong, J.O., Warsame, M., Agbenyega, T., Babiker, A.,Baiden, F., Yunus, E.B., Binka, F., Clerk, C., Folb, P., Hassan, R., Hossain, M.A.,Kimbute, O., Kitua, A., Krishna, S., Makasi, C., Mensah, N., Mrango, Z., Olliaro, P.,Peto, R., Peto, T.J., Rahman, M.R., Ribeiro, I., Samad, R., White, N.J., 2009. Study 13research group. Pre-referral rectal artesunate to prevent death and disability insevere malaria: a placebo-controlled trial. Lancet 373, 557–566.

Hoban, D.J., Doern, G.V., Fluit, A.C., Roussel-Delvallez, M., Jones, R.N., 2001.Worldwide prevalence of antimicrobial resistance in Streptococcus pneumoniae,Haemophilus influenzae, and Moraxella catarrhalis in the sentry antimicrobialsurveillance program, 1997–1999. Clinical Infectious Diseases 32, S81–S93.

Langtry, H.D., Balfour, J.A., 1998. Azithromycin: a review of its use in paediatricinfectious diseases. Drugs 56, 273–297.

Kauss, T., Gaubert, A., Ba, B., Boyer, C., Fawaz, F., Lembege, M., Boiron, J.M., Lafarge, X.,Lindegard, N., White, N.J., Olliaro, P.L., Millet, P., Gaudin, K., 2013. Pharmaceuticaldevelopment and optimization of azithromycin suppository for paediatric use.International Journal of Pharmaceutics 441, 218–226.

Kauss, T., Gaudin, K., Gaubert, A., Ba, B., Tagliaferri, S., Fawaz, F., Fabre, J.L., Boiron, J.M., Lafarge, X., White, N.J., Olliaro, P.L., Millet, P., 2012. Development of rectalforms of azithromycin as an alternative to oral and intra venous administrationfor paediatric treatment. International Journal of Pharmaceutics 436, 624–630.

Kauss, T., Fawaz, F., Guyot, M., Lagueny, A.M., Dos Santos, I., Bonini, F., Olliaro, P.,Millet, P., 2010. Fixed artesunate – amodiaquine combined pre-formulationstudy for the treatment of malaria. International Journal of Pharmaceutics 395,198–204.

Noedl, H., Krudsood, S., Leowattana, W., Tangpukdee, N., Thanachartwet, W.,Looareesuwan, S., Miller, R.S., Fukuda, M., Jongsakul, K., Yingyuen, K.,Sriwichhai, S., Ohrt, C., Knirsch, C., 2007. In vitro antimalarial activity ofazithromycin, artesunate and quinine in combination and correlation withclinical outcome. Antimicrobial Agents and Chemotherapy 51, 651–656.

Noedl, H., Wernsdorfer, H., Krudsood, S., Wilairatana, P., Kollaritsch, H., Wieder-mann, G., Looareesuwan, S., 2001. Antimalarial activity of azithromycin,artemisinin and dihydroartemisinin in fresh isolates of Plasmodium falciparumin Thailand. Acta Tropica 80, 39–44.

O’Dempsey, T.J., McArdle, T.F., Laurence, B.E., Lamont, A.C., Todd, J.E., Greenwood, B.M., 1993. Overlap in the clinical features of pneumonia and malaria in Africanchildren. Transactions of the Royal Society of Tropical Medicine and Hygiene 87,662–665.

O’Dempsey, T.J., McArdle, T.F., Lloyd-Evans, N., Laurence, B.E., Baldeh, I., Secka, O.,Greenwood, B.M., 1994. Importance of enteric bacteria as a cause of pneumonia,meningitis and septicemia among children in a rural community in The Gambia,West Africa. Pediatric Infectious Disease Journal 13, 122–128.

Pfizer, 2013. Pfizer’s Zithromax1 (Azithromycin) physician prescribing information,revised February 2013. Retrieved from http://labeling.pfizer.com/ShowLabel-ing.spx?id=511 (accessed in August 2013).

Rietveld, H.M., 1969. A profile refinement method for nuclear and magneticstructures. Journal of Applied Crystallography 2, 65–71.

Taylor, T.E., Fu, W.J., Carr, R.A., Whitten, R.O., Mueller, J.S., Fosiko, N.G., Lewallen, S.,Liomba, N.G., Molyneux, M.E., 2004. Differentiating the pathologies of cerebralmalaria by postmortem parasite counts. Nature Medicine 10, 143–145.

Tarning, J., Kloprogge, F., Piola, P., Dhorda, M., Muwanga, S., Turyakira, E.,Nuengchamnong, N., Nosten, F., Day, N.P., White, N.J., Guerin, P.J., Lindegardh,N., 2012. Population pharmacokinetics of artemether and dihydroartemisinin inpregnant women with uncomplicated plasmodium falciparum malaria inUganda. Malaria Journal 11, 293.

Whitty, C.J., Glasgow, K.W., Sadiq, S.T., Mabey, D.C., Bailey, R., 1999. Impact ofcommunity-based mass treatment for trachoma with oral azithromycin ongeneral morbidity in Gambian children. Pediatric Infectious Disease Journal 18,955–958.

WHO, 2010. Guidelines for the Treatment of Malaria, 2nd edition, World HealthOrganization. Retrieved from: http://whqlibdoc.who.int/publications/2010/9789241547925_eng.,df (accessed in August 2013).

WHO, 2012, WHO fact sheet N� 178, September 2012: available on http://www.who.int/mediacentre/factsheets/fs178/en/ (accessed in July 2013).

WHO, 2012, Update on artemisinin resistance – April 2012, available on http://www.who.int/malaria/publications/atoz/arupdate042012pdf?ua=1/ (accessedin January 2014).

Yeo, A.E., Rieckmann, K.H., 1995. Increased antimalarial activity of azithromycinduring prolonged exposure of Plasmodium falciparum in vitro. InternationalJournal for Parasitology 25, 531–532.